In this paper, a pretensioned pullout test method was proposed to better understand the bond mechanism and evaluate the bond behavior of pretensioned concrete members. A total of 15 pullout specimens with a centered 18‐mm (0.7 in.) diameter strand were tested. The specimens were divided into three sets according to their strand embedment lengths. For each set, there are two non‐pretensioned specimens and three pretensioned specimens with different initial prestress levels. In this way, the effects of the strand embedment length and the pretension level on the bond behavior were studied, in terms of the development of strand slip subject to the pullout force, initial P–S slope, the distribution of concrete surface strain, the ultimate failure mode and the pullout energy. The experimental results verified that the strand transfer length was dependent of prestress level. The transfer length may be predicted based on the different level of effective pretension force or stress in pretensioned specimens with different strand embedment lengths at detensioning. In essence, the disparities of the bond behaviors in the tests were caused by the contributions of different bond mechanisms. In return, the former, especially the characteristics of the development of strand slip and pullout energy, may be used for the quantificational evaluation of the bond mechanisms. Hoyer effect is verified as the dominant bond mechanism within transfer length.